AN UNSUPERVISED TEXTURE SEGMENTATION ALGORITHM WITH FEATURE SPACE REDUCTION AND KNOWLEDGE FEEDBACK

This paper presents an unsupervised texture segmentation algorithm bas ed on feature extraction using multichannel Gabor filtering, It is sho wn that feature contrast, a criterion derived for Gabor filter paramet er selection, is well suited for feature coordinate weighting in order to reduce the feature space dimension, The central idea of the propos ed segmentation algorithm is to decompose the actual segmented image i nto disjunct areas called scrap images and use them after lowpass filt ering as additional features for repeated k-means clustering and minim um distance classification, This yields a classification of texture re gions with an improved degree of homogeneity while preserving precise texture boundaries.